Development of a prognostic nomogram for long-term renal function in chronic kidney disease patients on the basis of...

To investigate the factors influencing the occurrence of intermediate and long-term renal failure in patients with chronic kidney disease (CKD), to construct a risk prediction model, and to provide a reference for the timing of substitution therapy in these patients.

A total of 354 patients who attended the Nephrology Department of Guizhou Medical University Hospital from January 2018 to July 2019 were used as the training set. The variables were screened using methods such as LASSO regression and the model was visualized using a nomogram. A total of 139 patients who attended the Department of Nephrology of the First People's Hospital of Bijie between November 2018 and June 2021 were used as an external validation cohort, and analyses were performed to verify the judgmental ability and clinical utility of the model.

Multifactorial Cox regression analysis revealed that hypertension, mesangial proliferation, renal tubular atrophy, serum potassium, and estimated glomerular filtration rate were independent risk factors for renal failure. The 1-year AUC values of internal validation and external validation were 0.978 (0.970-0.984) and 0.919 (0.853-0.986), respectively. The 2-year AUCs were 0.945 (0.914-0.961) and 0.902 (0.822-0.982), respectively. The AUC values over three years were 0.933 (0.897-0.945) and 0.915 (0.846-0.985), respectively. The bias correction curves were close to the ideal curves and revealed a high degree of consistency between actual observations and predictions.

We have developed a nomogram with excellent predictive power regarding the likelihood and time to progression to end-stage renal disease in CKD patients at 1, 2, and 3 years, which may help to classify patients and select the appropriate treatment plan in advance.